Bouncing towards the optimum: Improving the results of Monte Carlo optimization algorithms

DOI to cite this document:: 10.5283/epub.16106

Schneider, Johannes ; Morgenstern, Ingo ; Singer, Johannes Maria

PDF
(305kB) - Repository staff only

Date of publication of this fulltext: 09 Aug 2010 12:13

Alternative links to fulltext:DOI

Details

Export bibliographical data

Abstract

Simulated annealing and related Monte Carlo-type optimization algorithms are used to apply statistical physics concepts, in particular ideas from the statistical mechanics of spin glasses, to find optimal configurations for combinatorial optimization problems. There are formal proofs showing that these algorithms converge asymptotically (i.e.—possibly—for infinitely long simulation times) to a ...